When an AI Skill becomes paid, the hard part is not just adding a price field. The hard part is keeping the state machine honest.
These states should not be collapsed into one "published" or "paid" flag:
- package readiness;
- service readiness;
- a 402 / payment request;
- payment confirmation;
- fulfillment for the same order;
- platform review;
- settlement or actual revenue.
That is the reason we added a Tencent SkillHub Pay Adapter to TANCO SkillHub. The goal is to turn an ordinary Skill into an auditable Pay Skill candidate while preserving the original package, creating a clean converted copy, and recording package, service, payment, fulfillment, review, and settlement gates separately.
In the TANCO SkillHub stack, this also connects with other workflow Skills: Bianzhen checks evidence before an agent acts, Sunmao turns vague requirements into acceptance criteria, Lianzhu keeps complex tasks moving from judgment to execution to verification, and Huqiang keeps boundary and risk checks explicit.
Main TANCO SkillHub page:
https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94
Adapter:
https://skillhub.cn/team-skills/tencent-skillhub-pay-adapter
I’m curious how other agent builders handle this. Do you model payment and fulfillment as one tool call, or as separate workflow gates?
Top comments (0)